May 26, 2016 | Research News
By Luqiao Liu, Robert J. Shillman (1974) Career Development Assistant Professor in the Department of Electrical Engineering and Computer Science, MIT Microsystems Technology Laboratories
The high power consumption of computational electronic devices has become one of the major obstacles that prevents further improvement of their performance. Conventional electronic devices utilize electron charge to represent and store information. Therefore, Joule heating related energy dissipation always exists in these operations. A potential approach to reduce the power consumption is to use a spin-based device, in which the basic operation is conducted through the flipping of electrons’ spin rather than the motion of their charge. Spintronic devices also have advantages in non-volatility and scalability. Over the past decades, extensive studies have been carried out to utilize spintronic devices for memory, logic, and microwave applications.
Figure 1: (a) Schematic illustration of the spin transfer torque mechanism in a ferromagnet/nonmagnetic spacer/ferromagnet tri-layer structure. The open circles with arrows represent the spin orientation of conduction electrons while the bold arrows illustrate the magnetic moment orientation of the two ferromagnet layer. (b) Illustration of the spin generation from spin Hall effect. Arrows with different colors represent the spin orientations at different surfaces.
One of the key prerequisites to create low power spintronic devices is to find a mechanism that can switch electron spins efficiently. Conventionally, in order to write the information into a bit represented by electron spins, one has to apply a magnetic field. The magnetic field can usually be generated from a current-carrying coil. The necessity of a large dissipative charge current undercuts any energy gain through the transition from charge to spin. Moreover, the nonlocality of the current-induced magnetic field also made it impossible to scale the devices into small dimensions. About fifteen years ago, it was discovered that besides magnetic field, spin polarized charge currents can also be used to switch electron spins, or more precisely, the moment orientation of a nanomagnet[1,2]. In the tri-layer structure shown in Fig. 1(a), when conduction electrons go through a ferromagnetic electrode their spins will be aligned along the direction of the local ferromagnetic moment, due to the spin-spin interaction. If the moment orientations of the two ferromagnetic electrodes are not parallel with each other, the conduction electrons will transfer their spin angular momentum onto the local ion, when they go from one electrode towards the other. The discovery of this current-induced magnetic switching effect (also known as spin-transfer torque effect) made it possible to manipulate magnetic moments locally and efficiently. Based on this mechanism, a non-volatile memory — spin-transfer torque magnetic random access memory (STT-MRAM) — is currently being developed by many computer memory manufacturers.
While the spin-transfer torque effect provides a convenient way for controlling the magnetic moment orientation electrically, the limited conversion efficiency from a charge current into a spin current still makes it difficult to replace the existing devices (particularly memory devices) with the new ones. Recently we discovered that besides the tri-layer structure shown in Fig. 1(a), spin-transfer torque could also be realized in a much simpler system, where there are only one ferromagnetic layer and one non-magnetic layer[3,4]. Here, the generation of the spin-transfer torque is due to an effect known as “spin Hall effect,” which can be understood as a spin version of the well-known “Hall effect.” In the spin Hall effect, up and down spins, rather than positive and negative charges, get deflected towards different surfaces in the transverse direction as they move along the longitudinal direction (Figure 1(b)). More fundamentally, the deflection of different spins originates from the spin orbit interaction in solid crystals. In this non-magnetic layer, because of the spin Hall effect, excessive spins will accumulate at the interface when a longitudinal charge current is applied. Those accumulated spins will exert influence onto the ferromagnetic electrode that is in contact with them, which can further reorient the magnetic moment direction therein. With this spin Hall effect induced torque, we and other researchers have demonstrated that various types of magnetic dynamics such as magnetic switching[4-6], persistent magnetic oscillation[7,8], ferromagnetic resonance and magnetic domain wall motion[9,10] could be realized. In particular, a three terminal magnetic memory device is shown in Figure 2, where the switching of the magnetic moment is realized through the spin Hall effect and the reading is via the tunneling magnetoresistance effect.
Figure 2: (a) Schematic of device geometry used to switch magnetic moment of ferromagnet with spin Hall effect. (b) Spin Hall effect induced magnetic switching as is detected by the change in the magnetoresistance of the magnetic tunnel junction.
The biggest advantage associated with using spin orbit interaction to control the magnetic moment orientation lies in the fact that theoretically no upper limit exists for the charge current to spin current conversion efficiency. Empirically, one can use the “spin Hall angle” to quantify the efficiency, which is defined as the ratio between the spin current density 2e/ħ • JS that flows vertically into the FM electrode and the charge current density JC that is applied longitudinally. Here, e represents the electron charge and ℏ is Plank’s constant. In principle, using spin orbit interaction, the ratio between those two current densities can be far above one, while in contrast, this value can only go up to one in the spin filtering effect shown in Figure 1(a). In order to fully optimize the spin orbit interaction induced torque and to reduce the power consumption, a material with a large effective spin Hall angle is highly desirable. Recently it was realized that topological insulators could provide such capabilities[11-13]. Topological insulators are a new category of materials that were discovered only a few years ago. In their bulk, they exhibit insulating behaviors while on the surface, they behave like metals. Particularly, their surface states are spin polarized, meaning that a longitudinally flowing current will naturally lead to spin accumulation at the surfaces (see Figure 3(a)). Roughly, the topological insulators can be viewed as a quantum limit of the spin Hall effect discussed above, similar to the relationship between the quantum Hall effect and the ordinary Hall effect. Therefore, theoretically topological insulators would have the highest charge-to-spin conversion efficiency among the spin Hall effect in other material systems. Previously, we experimentally determined the spin current generation efficiency in two typical topological insulators, Bi2Se3 and (Bi,Sb)2Te3, using a spin polarized tunneling technique[14,15]. As is shown in Figure 3(b), it was discovered that several orders of magnitude improvement has been achieved in the effective spin Hall angle through the utilization of the topological insulators. This increase in the effective spin Hall angle indicates that the needed switching current can be lowered correspondingly. In the same figure, we also listed the corresponding resistivity of the studied materials. As the power consumption is given by I2 • R (Joule’s law), one can see that the decrease of the critical current in topological insulators is only partially cancelled by the increase of resistivity when calculating the power. A reduction of almost 600 times in energy dissipation is expected by going from heavy metals to topological insulators.
Figure 3: (a) Schematic illustration of the spin polarized surface states. (b) Effective spin Hall angle (blue square) and resistivity (green circles) of typical heavy metals and topological insulators obtained from spin polarized tunneling experiment. The data in this figure are from ref  and .
While it is possible to largely reduce the energy cost for magnetic switching by utilizing topological insulators, several difficulties still exist in directly combining topological insulators with a ferromagnetic electrode. One of the issues is the mismatch of impedance in those two materials. As one can tell from their names, topological insulators are in nature insulators or semiconductors, which have very high resistivity compared with ferromagnetic metals. Most of the current will be shunted through the ferromagnetic metal layers if an electrical voltage is applied across along the ferromagnetic metal/topological insulator bilayer film, which will dissipate large amount of energy. A ferromagnetic film which has similar impedance with topological insulator will be highly desirable to fully exploit the spin generation efficiency. Recently it was discovered that a magnetic oxide or antiferromagnetic oxide could be potentially used as a buffer layer between topological insulator and ferromagnetic metal. Those oxides, while being insulating to charge flow, allow spin currents to transmit through them with little resistance. By integrating this extra layer into the device shown in Figure 2, we believe that spintronic devices with ultralow power consumptions could be finally achievable.
 Slonczewski, J. C. Current-driven excitation of magnetic multilayers. Journal of Magnetism and Magnetic Materials 159, L1 (1996).
 Katine, J. A., Albert, F. J., Buhrman, R. A., Myers, E. B. & Ralph, D. C. Current-driven magnetization reversal and spin-wave excitations in Co/Cu/Co pillars. Phys. Rev. Lett. 84, 3149-3152 (2000).
 Liu, L., Moriyama, T., Ralph, D. & Buhrman, R. Spin-torque ferromagnetic resonance induced by the spin Hall effect. Phys. Rev. Lett. 106, 036601 (2011).
 Liu, L., Pai, C.-F., Li, Y., Tseng, H., Ralph, D. & Buhrman, R. Spin-torque switching with the giant spin Hall effect of tantalum. Science 336, 555-558 (2012).
 Liu, L., Lee, O., Gudmundsen, T., Ralph, D. & Buhrman, R. Current-induced switching of perpendicularly magnetized magnetic layers using spin torque from the spin Hall effect. Phys. Rev. Lett. 109, 096602 (2012).
 Miron, I. M. et al. Fast current-induced domain-wall motion controlled by the Rashba effect. Nature Mater. 10, 419-423 (2011).  Liu, L., Pai, C.-F., Ralph, D. & Buhrman, R. Magnetic oscillations driven by the spin hall effect in 3-terminal magnetic tunnel junction devices. Phys. Rev. Lett. 109, 186602 (2012).
 Demidov, V. E., Urazhdin, S., Ulrichs, H., Tiberkevich, V., Slavin, A., Baither, D., Schmitz, G. & Demokritov, S. O. Magnetic nano-oscillator driven by pure spin current. Nature Mater. 11, 1028-1031 (2012).
 Emori, S., Bauer, U., Ahn, S. M., Martinez, E. & Beach, G. S. Current-driven dynamics of chiral ferromagnetic domain walls. Nature Mater. 12, 611-616 (2013).
 Ryu, K.-S., Thomas, L., Yang, S.-H. & Parkin, S. Chiral spin torque at magnetic domain walls. Nature Nanotech 8, 527-533 (2013).  Mellnik, A. R. et al. Spin-transfer torque generated by a topological insulator. Nature 511, 449-451 (2014).
 Fan, Y. et al. Magnetization switching through giant spin–orbit torque in a magnetically doped topological insulator heterostructure. Nature Mater. 13, 699-704 (2014).
 Li, C. H., van `t Erve, O. M. J., Robinson, J. T., Liu, Y., Li, L. & Jonker, B. T. Electrical detection of charge-current-induced spin polarization due to spin-momentum locking in Bi2Se3. Nature Nanotech 9, 218-224 (2014).
 Liu, L., Richardella, A., Garate, I., Zhu, Y., Samarth, N. & Chen, C.-T. Spin-polarized tunneling study of spin-momentum locking in topological insulators. Phys. Rev. B 91, 235437 (2015).
 Liu, L., Chen, C.-T. & Sun, J. Z. Spin Hall effect tunnelling spectroscopy. Nature Phys. 10, 561-566 (2014).
 Wang, H., Du, C., Hammel, P. C. & Yang, F. Antiferromagnonic Spin Transport fromY3Fe5O12 into NiO. Phys. Rev. Lett. 113, 097202 (2014).